Memory-mapped file

Memory-mapped file, as its name suggest, maps address space in RAM to a file that is accessible and visible via the manipulation of the file system on your Operating System. To understand the usage of the mmap() system call, the following illustration will help explains the concept of sharing flag while using mmap():

void *mmap(void *addr, size_t length, int prot, int flags, int fd, off_t offset);

In order to use mmap(), we need to know the size of the memory we are allocating and we need to use open() to create a file descriptor for mmap() to associate memory address space in the physical RAM to an actual file in your file system under the Operating System.

mmap() also requires permission specification by setting the PROT flag, there are quite a few permission we can set, PROT_READ, PROT_WRITE, PROT_EXEC and PROT_NONE. In the end, we can also specify an offset for the contents of a file mapping to begin at certain length inside the file, an offset must be a multiple of the page size as returned by sysconf(_SC_PAGE_SIZE).

#include <stdio.h>
#include <fcntl.h>
#include <unistd.h>
#include <sys/mman.h>

int main(int argc, char **argv)
{
    int fd, sharing_flags, permission;
    off_t offset = 0;
    size_t s = sizeof(int*8);
    int* addr;

    fd = open(argv[1], O_RDONLY);

    permission = PROT_READ|PROT_WRITE;
    sharing_flags = MAP_PRIVATE;

    addr = (int *) mmap(NULL, s, permission, sharing_flags, fd, offset);

    if(addr == MAP_FAILED)
    {
        perror("Error while allocating address space with mmap()");
        exit(1);
    }

    *(addr[0]) = 234;
    close(fd);

}

After we are done with the mapped file, we need to call munmap() to disassociate the mapped file from physical RAM, the usage is as follow:

{
...
    if(munmap(addr, s) == -1) // We need to perform error validation in order to know if the operation is performed correctly.
    {
        perror("Error while unmapping the mapped file from RAM");
        exit(1);
    }
...
}